Colorbar on Geopandas

Question

I am trying to create a Matplotlib colorbar on GeoPandas.

import geopandas as gp
import pandas as pd
import matplotlib.pyplot as plt

#Import csv data
df = df.from_csv('data.csv')

#Convert Pandas DataFrame to GeoPandas DataFrame
g_df = g.GeoDataFrame(df)

#Plot
plt.figure(figsize=(15,15)) 
g_plot = g_df.plot(column='column_name',colormap='hot',alpha=0.08)
plt.colorbar(g_plot)

I get the following error:

AttributeError                            Traceback (most recent call last)
<ipython-input-55-5f33ecf73ac9> in <module>()
      2 plt.figure(figsize=(15,15))
      3 g_plot = g_df.plot(column = 'column_name', colormap='hot', alpha=0.08)
----> 4 plt.colorbar(g_plot)

...

AttributeError: 'AxesSubplot' object has no attribute 'autoscale_None'

I am not sure how to get colorbar to work.

Answer 1

EDIT: The PR referenced below has been merged into the geopandas master. Now you can simply do:

gdf.plot(column='val', cmap='hot', legend=True)

and the colorbar will be added automatically.

Notes:

• legend=True tells Geopandas to add the colorbar.
• colormap is now called cmap.
• vmin and vmax are not required anymore.

See https://geopandas.readthedocs.io/en/latest/mapping.html#creating-a-legend for more (with an example how to adapt the size and placement of the colorbar).

---

There is a PR to add this to geoapandas (https://github.com/geopandas/geopandas/pull/172), but for now, you can add it yourself with this workaround:

## make up some random data
df = pd.DataFrame(np.random.randn(20,3), columns=['x', 'y', 'val'])
df['geometry'] = df.apply(lambda row: shapely.geometry.Point(row.x, row.y), axis=1)
gdf = gpd.GeoDataFrame(df)

## the plotting

vmin, vmax = -1, 1

ax = gdf.plot(column='val', colormap='hot', vmin=vmin, vmax=vmax)

# add colorbar
fig = ax.get_figure()
cax = fig.add_axes([0.9, 0.1, 0.03, 0.8])
sm = plt.cm.ScalarMappable(cmap='hot', norm=plt.Normalize(vmin=vmin, vmax=vmax))
# fake up the array of the scalar mappable. Urgh...
sm._A = []
fig.colorbar(sm, cax=cax)

The workaround comes from Matplotlib - add colorbar to a sequence of line plots. And the reason that you have to supply vmin and vmax yourself is because the colorbar is not added based on the data itself, therefore you have to instruct what the link between values and color should be.